Page 18 - Docker

A container that runs fine on your laptop is not yet a production-ready container. Health checks, restart policies, secrets management, resource limits and clean logging are the factors that decide between a stable production system and outages in the middle of the night.

Self-signed certificates constantly trigger browser warnings and can't be used for APIs or service workers. Combining mkcert with a Docker reverse proxy solves the problem completely: local domains with real, trusted HTTPS, no cloud, no Let's Encrypt, no endless clicking.

When a monorepo combines several largely independent sub-projects in a single repository, one monolithic .gitlab-ci.yml quickly becomes confusing and inefficient. Parent-child pipelines with the trigger keyword solve this by giving every sub-project its own, independently executable child pipeline that only runs when something in its area has actually changed.

LXC provides true system containers with a full userspace, its own init system and several parallel services, while Docker packages a single application process. For admins looking for a lightweight alternative to virtual machines, the LXD workflow, from installation through networking and storage to snapshots, offers a practical foundation for production environments where a fully-fledged operating system is needed rather than a single process.

Anyone who switches to the terminal for every docker logs call loses time and context. PhpStorm ships with a complete Docker integration: log streaming, service control, process inspection and a remote interpreter, all without ever leaving the IDE.

Docker containers usually run fine in testing, but in continuous operation log rotation, the right storage driver, clean systemd integration and clear resource limits decide whether a single container brings down the whole server or keeps running quietly in the background. This article shows how administrators configure the Docker daemon for production, keep containers starting reliably and continuously monitor their health status.

Teams looking for a Docker Desktop alternative because of licensing rules or resource usage almost always end up comparing Colima and Podman. Both replace the commercial Docker Desktop daemon with leaner, open solutions, but they differ significantly in architecture, compatibility and everyday usability for development teams.

Docker Scout is an analysis tool built into Docker Desktop and the Docker CLI that surfaces CVEs in images, gives recommendations for leaner base images, and fits into the daily workflow without installing a separate tool.

Seed data is fundamentally different from plain init scripts: init scripts create the schema, seed data fills it with realistic, repeatable test data. Anyone mixing up both concepts ends up either with empty development databases or with inconsistent test states between team members.

Credentials in the environment section of a compose.yaml end up in docker inspect, in process lists and often accidentally in log output. The secrets block in Docker Compose solves this problem by delivering credentials as files inside the container instead of as visible environment variables, creating a significantly cleaner way of handling passwords and API keys.